skill-creator

Name: skill-creator
Author: sabahattink

27stars9forks

How to install skill-creator?

sabahattink/antigravity-fullstack-hq/skill-creator

$npx -y skills add sabahattink/antigravity-fullstack-hq --skill skill-creator

Installs into the current project.

Summary

How to create new skills for this HQ — format, frontmatter, content structure. Use when you need to add a new skill to this repository, or when reviewing whether an existing skill is well-formed.

Files · 1

View on GitHub

SKILL.md

1	# Skill Creator
2
3	## What is a Skill?
4
5	A skill is a reference document that gives an AI agent (or a developer) deep, actionable knowledge for a specific domain. It is not a README, a tutorial, or a one-liner tip. It contains real patterns, real code, and clear guidance on what NOT to do.
6
7	## File Location
8
9	```
10	skills/
11	└── <skill-name>/
12	└── SKILL.md
13	```
14
15	Skill names are lowercase, hyphenated: `api-design-patterns`, `auth-patterns`, `docker-patterns`.
16
17	## Required Structure
18
19	```markdown
20	---
21	name: skill-name
22	description: One-line description. Use when [specific trigger condition].
23	---
24
25	# Title
26
27	## [Section 1]
28	[Content with real code examples]
29
30	## [Section 2]
31	...
32
33	## Forbidden Patterns
34	[What NOT to do — these are as important as the positive guidance]
35	```
36
37	## Frontmatter Rules
38
39	```yaml
40	---
41	name: skill-name # must match the directory name
42	description: > # one line: what it does + when to trigger
43	Generates Excel files with ExcelJS. Use when creating .xlsx exports
44	from database data in a NestJS app.
45	---
46	```
47
48	The `description` field is what an AI uses to decide whether to load this skill. Make the trigger condition specific:
49
50	```
51	# Good — specific trigger
52	description: JWT access/refresh tokens, Passport.js strategies. Use when implementing auth in a NestJS + Next.js app.
53
54	# Bad — too vague
55	description: Authentication patterns. Use when building auth.
56	```
57
58	## Content Rules
59
60	### 1. Real code, not pseudocode
61	Every code block must be runnable with the imports shown. Never use `...` to hide required setup.
62
63	```typescript
64	// Good — complete, runnable snippet
65	import { Injectable } from '@nestjs/common'
66	import { InjectRepository } from '@nestjs/typeorm'
67	import { Repository } from 'typeorm'
68	import { User } from './user.entity'
69
70	@Injectable()
71	export class UsersRepository {
72	constructor(
73	@InjectRepository(User)
74	private readonly repo: Repository<User>
75	) {}
76
77	findById(id: number) {
78	return this.repo.findOne({ where: { id } })
79	}
80	}
81
82	// Bad — too abstract
83	class MyRepo {
84	// inject the repo here
85	findById(id) { ... }
86	}
87	```
88
89	### 2. Forbidden Patterns section is mandatory
90	The "Forbidden Patterns" section must be the last section. List 5-10 concrete things that cause real bugs, security issues, or maintenance problems.
91
92	```markdown
93	## Forbidden Patterns
94
95	- Never store access tokens in localStorage — XSS can steal them
96	- Never use the same JWT secret for access and refresh tokens
97	- Never skip `@IsEmail()` validation on email fields
98	```
99
100	### 3. Aim for 200-400 lines
101	- Under 200 lines: too shallow, not useful as a reference
102	- Over 400 lines: likely covering too many topics, should be split
103
104	### 4. One topic per skill
105	Don't mix unrelated concerns. If you're writing a `testing` skill, don't include deployment. If the topic is naturally broad, split it:
106
107	```
108	# Too broad:
109	skills/backend/SKILL.md ← covers NestJS + TypeORM + auth + caching
110
111	# Better:
112	skills/backend-dev-guidelines/SKILL.md ← NestJS layering
113	skills/auth-patterns/SKILL.md ← JWT + Passport
114	skills/docker-patterns/SKILL.md ← Containerization
115	```
116
117	## Section Patterns to Follow
118
119	### Command blocks for setup
120	```bash
121	npm install exceljs
122	npm install -D @types/node
123	```
124
125	### Configuration files
126	Show actual config files, not fragments.
127
128	### Patterns with comparison (good vs bad)
129	```typescript
130	// Bad: explanation of why it's bad
131	const bad = doWrongThing()
132
133	// Good: explanation of why it's better
134	const good = doRightThing()
135	```
136
137	### Checklists for pre-flight checks
138	```
139	Production Deployment Checklist:
140	□ NODE_ENV=production set
141	□ Secrets validated at startup
142	□ Health checks responding
143	□ Rate limiting enabled
144	```
145
146	## Quality Checklist
147
148	Before merging a new skill:
149
150	```
151	□ Frontmatter has name + description
152	□ name matches directory name
153	□ Description includes "Use when [specific condition]"
154	□ Code examples are complete (have imports, no ellipsis hiding setup)
155	□ At least 3 major sections with real content
156	□ "Forbidden Patterns" section exists and has 5+ entries
157	□ 200-400 lines total
158	□ No tutorial prose — reference-style only
159	□ TypeScript/language specifics match the project stack
160	```
161
162	## Example: Creating a New Skill
163
164	```bash
165	# 1. Create the directory and file
166	mkdir skills/redis-patterns
167	touch skills/redis-patterns/SKILL.md
168
169	# 2. Write the skill following the template above
170
171	# 3. Commit
172	git add skills/redis-patterns/SKILL.md
173	git commit -m "feat: add redis-patterns skill"
174	```
175
176	## Forbidden Patterns
177
178	- Never write a skill that's just a list of links — content must be here, not linked
179	- Never use placeholders like `[your code here]` — write the actual code
180	- Never mix two unrelated domains in one skill — split instead
181	- Never skip the "Forbidden Patterns" section — negative examples are the most

People who liked this also viewed

6 picks

Type

skill-creator

27stars9forks

View on GitHub

How to install skill-creator?

sabahattink/antigravity-fullstack-hq/skill-creator

$npx -y skills add sabahattink/antigravity-fullstack-hq --skill skill-creator

Installs into the current project.

Summary

How to create new skills for this HQ — format, frontmatter, content structure. Use when you need to add a new skill to this repository, or when reviewing whether an existing skill is well-formed.

Files · 1

View on GitHub

SKILL.md

1	# Skill Creator
2
3	## What is a Skill?
4
5	A skill is a reference document that gives an AI agent (or a developer) deep, actionable knowledge for a specific domain. It is not a README, a tutorial, or a one-liner tip. It contains real patterns, real code, and clear guidance on what NOT to do.
6
7	## File Location
8
9	```
10	skills/
11	└── <skill-name>/
12	└── SKILL.md
13	```
14
15	Skill names are lowercase, hyphenated: `api-design-patterns`, `auth-patterns`, `docker-patterns`.
16
17	## Required Structure
18
19	```markdown
20	---
21	name: skill-name
22	description: One-line description. Use when [specific trigger condition].
23	---
24
25	# Title
26
27	## [Section 1]
28	[Content with real code examples]
29
30	## [Section 2]
31	...
32
33	## Forbidden Patterns
34	[What NOT to do — these are as important as the positive guidance]
35	```
36
37	## Frontmatter Rules
38
39	```yaml
40	---
41	name: skill-name # must match the directory name
42	description: > # one line: what it does + when to trigger
43	Generates Excel files with ExcelJS. Use when creating .xlsx exports
44	from database data in a NestJS app.
45	---
46	```
47
48	The `description` field is what an AI uses to decide whether to load this skill. Make the trigger condition specific:
49
50	```
51	# Good — specific trigger
52	description: JWT access/refresh tokens, Passport.js strategies. Use when implementing auth in a NestJS + Next.js app.
53
54	# Bad — too vague
55	description: Authentication patterns. Use when building auth.
56	```
57
58	## Content Rules
59
60	### 1. Real code, not pseudocode
61	Every code block must be runnable with the imports shown. Never use `...` to hide required setup.
62
63	```typescript
64	// Good — complete, runnable snippet
65	import { Injectable } from '@nestjs/common'
66	import { InjectRepository } from '@nestjs/typeorm'
67	import { Repository } from 'typeorm'
68	import { User } from './user.entity'
69
70	@Injectable()
71	export class UsersRepository {
72	constructor(
73	@InjectRepository(User)
74	private readonly repo: Repository<User>
75	) {}
76
77	findById(id: number) {
78	return this.repo.findOne({ where: { id } })
79	}
80	}
81
82	// Bad — too abstract
83	class MyRepo {
84	// inject the repo here
85	findById(id) { ... }
86	}
87	```
88
89	### 2. Forbidden Patterns section is mandatory
90	The "Forbidden Patterns" section must be the last section. List 5-10 concrete things that cause real bugs, security issues, or maintenance problems.
91
92	```markdown
93	## Forbidden Patterns
94
95	- Never store access tokens in localStorage — XSS can steal them
96	- Never use the same JWT secret for access and refresh tokens
97	- Never skip `@IsEmail()` validation on email fields
98	```
99
100	### 3. Aim for 200-400 lines
101	- Under 200 lines: too shallow, not useful as a reference
102	- Over 400 lines: likely covering too many topics, should be split
103
104	### 4. One topic per skill
105	Don't mix unrelated concerns. If you're writing a `testing` skill, don't include deployment. If the topic is naturally broad, split it:
106
107	```
108	# Too broad:
109	skills/backend/SKILL.md ← covers NestJS + TypeORM + auth + caching
110
111	# Better:
112	skills/backend-dev-guidelines/SKILL.md ← NestJS layering
113	skills/auth-patterns/SKILL.md ← JWT + Passport
114	skills/docker-patterns/SKILL.md ← Containerization
115	```
116
117	## Section Patterns to Follow
118
119	### Command blocks for setup
120	```bash
121	npm install exceljs
122	npm install -D @types/node
123	```
124
125	### Configuration files
126	Show actual config files, not fragments.
127
128	### Patterns with comparison (good vs bad)
129	```typescript
130	// Bad: explanation of why it's bad
131	const bad = doWrongThing()
132
133	// Good: explanation of why it's better
134	const good = doRightThing()
135	```
136
137	### Checklists for pre-flight checks
138	```
139	Production Deployment Checklist:
140	□ NODE_ENV=production set
141	□ Secrets validated at startup
142	□ Health checks responding
143	□ Rate limiting enabled
144	```
145
146	## Quality Checklist
147
148	Before merging a new skill:
149
150	```
151	□ Frontmatter has name + description
152	□ name matches directory name
153	□ Description includes "Use when [specific condition]"
154	□ Code examples are complete (have imports, no ellipsis hiding setup)
155	□ At least 3 major sections with real content
156	□ "Forbidden Patterns" section exists and has 5+ entries
157	□ 200-400 lines total
158	□ No tutorial prose — reference-style only
159	□ TypeScript/language specifics match the project stack
160	```
161
162	## Example: Creating a New Skill
163
164	```bash
165	# 1. Create the directory and file
166	mkdir skills/redis-patterns
167	touch skills/redis-patterns/SKILL.md
168
169	# 2. Write the skill following the template above
170
171	# 3. Commit
172	git add skills/redis-patterns/SKILL.md
173	git commit -m "feat: add redis-patterns skill"
174	```
175
176	## Forbidden Patterns
177
178	- Never write a skill that's just a list of links — content must be here, not linked
179	- Never use placeholders like `[your code here]` — write the actual code
180	- Never mix two unrelated domains in one skill — split instead
181	- Never skip the "Forbidden Patterns" section — negative examples are the most

People who liked this also viewed

6 picks

Type